We propose to develop, using genes specifying essential proteins of the yeast Saccharomyces cerevisiae, a way to study protein structure/function relationships in the living cell. The method, which involves a new generation of mutagenesis strategies, will be applied to yeast proteins that are highly homologous (and functionally interchangeable) with a human counterpart and for which a highresolution X-ray structure is available. We plan to begin this study with the single yeast actin gene (ACT1) and later extend it to one or more members of the small GTP-binding or RAS oncogene superfamily. The method consists of three elements. The first (already accomplished in the case of actin) is a comprehensive "charged-toalanine" mutagenesis-scan of protein surface regions for mutations that allow synthesis and folding but which prevent function as judged by the ability to support the growth of a haploid yeast cell in vivo. The second is analysis by "random-replacement mutagenesis" of the regions indicated by the scan to be essential for function. The third is molecular modeling of a representative sample of the variants in each essential region that are consistent with function. Random replacement mutagenesis and molecular modeling of functional replacement sequences has been applied already to the secreted enzyme TEM-beta-lactamase, a major determinant of plasmid-borne penicillin resistance in bacteria. The information gained from this kind of comprehensive mutagenic study should allow the determination of which parts of each of these proteins contact their ligands. In each case, the new mutations generated should facilitate isolation of interacting genes and proteins by standard genetic and biochemical methods that connect specifically to different-essential parts of the protein surface.